Factor H–IgG Chimeric Proteins as a Therapeutic Approach against the Gram-Positive Bacterial Pathogen Streptococcus pyogenes

This article is featured in In This Issue, p.3713

Streptococcus pyogenes, also known as group A Streptococcus, is a Gram-positive bacterium that causes a wide spectrum of diseases. Symptoms range from mild superficial skin infections to invasive and life-threatening disease (1, 2). Worldwide >700 million S. pyogenes infections occur annually with at least 663,000 new cases of invasive infections (3). To evade the immune system, S. pyogenes binds a variety of serum proteins, including albumin, fibronectin, plasminogen, IgG, complement factor H (FH), and C4b-binding protein (C4BP) (4–10). S. pyogenes is exclusively a human pathogen, which may in part be related to its ability to inhibit human complement by binding to the complement inhibitors FH and C4BP in a human-specific manner (4, 11, 12). Recently, we showed that recruitment of these human complement inhibitors is crucial for S. pyogenes virulence in the mouse model (13).

The complement system is a pivotal branch of innate immunity against invading pathogens by marking them for removal following opsonization with C3b (14). Complement is activated by surface-bound Abs and certain pathogen-associated molecular patterns, which initiate the protein cascade to eliminate bacteria, viruses, fungi, and non-self cells (15). Complement activation can occur via three pathways, namely the classical, lectin, and alternative pathways. All three pathways converge at the level of the C3 convertase, resulting in C3b deposition on the target (16). In addition to C3b deposition that promotes opsonophagocytosis, complement activation also releases anaphylatoxins such as C3a and C5a to stimulate the immune system and initiate a response. The final step of complement activation results in formation of a lytic pore, called the membrane attack complex, which can directly lyse Gram-negative bacteria and eukaryotic cells (17). Therefore, this powerful, but potentially detrimental cascade has to be tightly regulated to prevent unwanted host cell damage (16). In addition to several surface-bound complement regulators, C4BP and FH are two of the major soluble complement inhibitors, which protect the body’s own cells from unwanted complement activation and killing (18, 19).

Many bacteria have developed evasion strategies to prevent immune recognition. One strategy, also used by S. pyogenes as discussed above, is to recruit the host’s complement regulatory proteins FH and C4BP to the bacterial surface (20, 21). Ideally, these two complement inhibitors should exclusively protect host cells, but not microbes from complement attack (22). However, bacteria “hijack” these C4BP and FH to their surface (23) to escape opsonization and persist in the host (24). For example, several bacteria, including S. pyogenes (25), bind FH via domains 6–7 or 18–20 (reviewed in Ref. 26). Because the complement-inhibiting activity of FH is located in domains 1–4, FH bound to surfaces via domains 6–7 or 18–20 can effectively limit complement activation (27).

Since the identification of penicillin nearly 90 y ago (28), infectious diseases have been treated with various antibiotics. Unfortunately, more and more bacteria exhibit resistance against common antibiotics, and frequently against several antibiotic classes simultaneously (29, 30), which may herald a “postantibiotic era.” New approaches for antibacterial therapies are urgently needed (31).

To achieve this, we designed a chimeric protein consisting of domains 6 and 7 of human FH, which are fused to the Fc domain of human IgG1 (FH6-7/human Fc [hFc]). Many pathogens recruit human FH via domains 6 and 7 to inhibit complement activation on their surfaces (32). FH6-7/hFc bound to a pathogen is expected to compete out serum FH from the microbial surface, which would permit uninhibited alternative pathway amplification. Furthermore, the IgG1 Fc domain of the chimeric protein will activate the classical pathway of complement through interaction with C1q of the C1 complex. Finally, surface-bound FH6-7/hFc also can be recognized by Fc receptors (FcγR) on professional phagocytes to initiate opsonophagocytosis.

We have previously provided proof-of-concept of efficacy of FH/Fc molecules in the treatment of Gram-negative bacterial infections caused by Haemophilus influenzae, Neisseria meningitidis, and Neisseria gonorrhoeae (33–35). In this study, we examine the efficacy of human FH-Fc chimeric proteins on Gram-positive bacterial infections using the example of S. pyogenes.

S. pyogenes strains AP1 (S. pyogenes strain 40/58, serotype M1) and AP18 (strain 8/69, serotype M18), both from the World Health Organization Collaborating Center for Reference and Research on Streptococci (Prague, Czech Republic), MC25 (lacks M1 protein) (36), BM27.6 (lacks protein H and M1) (37), and BMJ71 (lacks protein H and M1, SIC, and C5a peptidase) (38) were grown in Todd–Hewitt broth medium (Oxoid) overnight at 37°C, 5% CO₂ without shaking. Mutant strains were cultured in medium supplemented with 150 μg/ml kanamycin (MC25), 1 μg/ml erythromycin (BM27.6), or 5 μg/ml tetracycline (BMJ71). Overnight bacteria cultures were diluted to OD₆₀₀ of 0.1 in fresh medium and grown with same conditions until exponential phase growth was reached at OD₆₀₀ of 0.3–0.4. After centrifugation and washing with 1× PBS, bacteria were ready to use.

FreeStyle Chinese hamster ovary (CHO)-S suspension cells (Life Technologies) were used for expression of fusion proteins. Cells were cultured in FreeStyle CHO expression medium (Thermo Fisher) supplemented with 8 mM l-glutamine (HyClone) at 37°C, 8% CO₂, 130 rpm shaking.

Protein M1, protein H, and its fragments were expressed in Escherichia coli and purified on a human IgG column as described previously (5, 37, 39). FH was purified from human plasma. Aggregated human IgG was obtained by heat aggregation of Igs (Kiovig, Baxalta) at 65°C. Zymosan and cytochalasin D were purchased from Sigma-Aldrich. FH and FH6-7/mouse Fc (mFc) fusion protein were radioactively labeled using Iodo-Beads (Pierce) and Na-¹²⁵I (PerkinElmer Life Sciences) according to the manufacturers’ protocols. Abs used for flow cytometric analysis included goat anti-human C9 (CompTech), donkey F(ab′)₂ anti-goat IgG conjugated to Alexa Fluor 647 (Jackson ImmunoResearch Laboratories), MRC OX-24 anti-human FH conjugated to DyLight 650 (40), goat anti-mouse IgG conjugated to Alexa Fluor 647 (Invitrogen), goat anti-human IgG conjugated to Alexa Fluor 647 (Invitrogen), and goat anti-human IgG conjugated to Alexa Fluor 488 (Invitrogen). Abs used for complement deposition assay included rabbit anti-human C4c (Dako), rabbit anti-human C3d (Dako), and goat anti-rabbit IgG conjugated to HRP (Dako). The following Abs used for phagocytosis assay were purchased from BioLegend (unless stated otherwise): anti-human CD56 conjugated to allophycocyanin/Cy7, anti-human CD15 conjugated to PerCP/Cy5.5, anti-human CD19 conjugated to allophycocyanin/Cy7, anti-human MHC class II conjugated to Alexa Fluor 647, anti-human CD11c conjugated to PE, anti-human CD3 conjugated to allophycocyanin/Cy7, anti-human CD16 conjugated to PE/Cy7, anti-human CD14 conjugated to PE/Dazzle, human TruStain FcX, and mouse anti–S. pyogenes (AbD Serotec) conjugated to Alexa Fluor 488 and BV405. Ten micrograms per milliliter of the following Abs was used to block Fcγ receptors (41): anti-human CD16 (no. 16-0166-85; Invitrogen), anti-human CD32 (no. LS-C187457; LSBio), and anti-human CD64 (no. 16-0649-85; eBioscience). Compstatin CP40, which blocks C3 activation, was prepared at the University of Pennsylvania as described previously (42). Abs for Western blot analysis included anti-human C3a/C3a desArg (Hycult Biotech) and anti-mouse Ig/HRP (Dako).

Chimeric proteins (FH6-7/18-20hIgG1, FH6-7/18-20mIgG2a) were expressed in CHO cells and purified on protein A/G columns as described previously (34). Briefly, floating FreeStyle CHO-S cells (Life Technologies) at a concentration of 1 × 10⁶ cells/ml were transfected with plasmids encoding fusion protein constructs in FreeStyle CHO expression medium (Thermo Fisher). Transfected floating FreeStyle CHO-S cells were incubated at 37°C and 8% CO₂ under shaking at 125–130 rpm for up to 12 d. Every second day cell medium containing the produced chimeric proteins was collected and stored at −20°C for subsequent protein purification. Then, cells were resuspended in fresh FreeStyle CHO expression medium for further incubation. Proteins within the collected cell supernatant were purified on a protein A/G column (GE Healthcare) at 4°C. The elution of the proteins was performed using glycine and 6 M guanidinium chloride. Eluates were transferred into Spectra pore membranes for dialysis in 1× PBS at 4°C with three buffer exchanges during 24 h. Protein concentration was determined using a Cary 50 Bio photometer (Varian).

Purified proteins (protein M1, protein H, protein H fragments, and α-1 antitrypsin [α1AT]) were diluted in HEPES buffer (50 mM HEPES, 50 mM NaCl) and immobilized in microtiter plates (MaxiSorp BreakApart; Nunc) overnight at 4°C. Next day, plates were washed three times with washing buffer (50 mM Tris [pH 8], 150 mM NaCl, 0.1% Tween 20) and unspecific binding was blocked with 5% Difco skim milk (Becton Dickinson) diluted in wash buffer. ¹²⁵I-FH and ¹²⁵I-FH6-7/mFc were diluted in HEPES buffer and added in increasing amounts as indicated. After overnight incubation at 4°C, plates were washed using washing buffer and radioactivity was measured using a Wizard² gamma counter (PerkinElmer Life Sciences).

Harvested bacteria S. pyogenes AP1 or BM27.6 were incubated with 10% normal human serum and increasing amounts of FH6-7/hFc or FH18-20/hFc, respectively, diluted in GVB⁺⁺ buffer (5 mM veronal buffer [pH 7.3], 140 mM NaCl, 0.1% gelatin, 1 mM MgCl₂, and 5 mM CaCl₂) for 1 h at 37°C, 5% CO₂. After incubation, bacteria were washed with 0.5% BSA in 1× PBS and stained for C9 (goat anti-human C9; CompTech) at 4°C for 30 min. After washing with 1× PBS, primary Abs were detected with secondary anti-goat Abs conjugated to Alexa Fluor 647 (Jackson ImmunoResearch Laboratories). The amount of deposited C9 component bound to bacteria surface was measured using CyFlow space flow cytometer (Partec).

Microtiter plates (MaxiSorp BreakApart; Nunc) were coated with FH6-7/hFc (10 μg/ml), α1AT (10 μg/ml), and aggregated human IgG (classical pathway, 5 μg/ml; Kiovig, Baxalta) or zymosan (alternative pathway, 10 μg/ml, Z-4250; Sigma-Aldrich). The plates were incubated overnight at 4°C and washed the next day three times with washing buffer (50 mM Tris [pH 8], 150 mM NaCl, 0.1% Tween 20). Between every step, plates were washed thrice with washing buffer. Further unspecific binding was blocked with 5% Difco skim milk (Becton Dickinson) diluted in wash buffer (classical pathway) or quenching buffer (3% fish gelatin, 50 mM Tris-HCl, 150 mM NaCl, 0.1% Tween 20 [pH 8]; alternative pathway). The next plate was incubated with increasing concentrations of normal human serum diluted in GVB⁺⁺ (5 mM veronal buffer [pH 7.3], 140 mM NaCl, 0.1% gelatin, 1 mM MgCl₂, and 5 mM CaCl₂; classical pathway) or Mg²⁺ EGTA (2.5 mM veronal buffer [pH 7.3] containing 70 mM NaCl, 140 mM glucose, 0.1% gelatin, 7 mM MgCl₂, and 10 mM EGTA; alternative pathway). Serum incubation time was 20 min for C3b, 15 min for C4b (classical pathway), and 30 min for C3b (alternative pathway) at 37°C. After incubation, deposited components were detected with primary anti-C3d (Dako) (classical pathway, alternative pathway) and anti-C4d (Dako) (classical pathway) diluted in corresponding blocking buffer. Then, bound primary Abs were detected with secondary anti-rabbit HRP-conjugated Abs. Finally, samples were developed using OPD tablets (Kem-En-Tec Diagnostics) and signal was measured at OD₄₉₀ (Cary 50 MPR microplate reader; Varian).

For classical pathway activation, 150 μl of sheep erythrocytes (Håtunalab; corresponding to ∼3.75 × 10⁸ RBCs) was washed three times with GVB⁺⁺ buffer (5 mM veronal buffer [pH 7.3], 140 mM NaCl, 0.1% gelatin, 1 mM MgCl₂, and 5 mM CaCl₂) and incubated with Amboceptor (1:500 in GVB⁺⁺; Dade Behring) for 20 min at 37°C with 650 rpm shaking. After incubation, erythrocytes were washed three times and resuspended in GVB⁺⁺. The suspension volume was adjusted in the way that 10 μl from erythrocyte stock lysed in 90 of H₂O resulted in OD₄₁₀ of 1–1.2. For the alternative pathway, 150 μl of rabbit erythrocytes (Håtunalab) was washed three times and suspended in Mg-EGTA buffer (2.5 mM veronal buffer [pH 7.3] containing 70 mM NaCl, 140 mM glucose, 0.1% gelatin, 7 mM MgCl₂, and 10 mM EGTA). Then, the suspension volume was adjusted in the same way as for sheep erythrocytes. To assess hemolytic activity, sheep erythrocytes were incubated with increasing amounts of FH6-7/hFc diluted in GVB⁺⁺ buffer for 20 min. After incubation, cells were washed and subsequently mixed with 0.75% normal human serum (NHS) to allow complement activation for 10 min. Remaining erythrocytes were pelleted and hemolysis was measured at OD₄₁₀ (Cary 50 MPR microplate reader; Varian). Similarly, rabbit erythrocytes were incubated with increasing amounts of FH6-7/hFc fusion protein diluted in Mg-EGTA buffer. After washing and mixing with 6% NHS, complement was allowed to be activated on erythrocytes for 15 min. Finally, erythrocytes were centrifuged and released hemoglobin was measured at OD₄₁₀.

All incubations were performed at 37°C and 650 rpm shaking (ThermoMixer; Eppendorf). Hemolytic assay using anti-CD59–treated human erythrocytes has been performed as described previously (35).

S. pyogenes strains AP1, MC25, BM27.6, and BMJ71 (1 × 10⁷ CFU) in GVB⁺⁺ were coincubated with indicated concentrations of chimeric FH proteins in the presence of 10% FCS at 37°C, 5% CO₂ for 1 h. Bacteria were harvested and washed with 1× PBS. Subsequently, bacteria were stained with CellTrace Calcein Violet (Molecular Probes) and stained with goat anti-human IgG–Alexa Fluor 488 (Invitrogen; to detect FH–human IgG1 proteins) or goat anti-mouse IgG–Alexa Fluor 647 (Invitrogen; to detect FH–mouse IgG2a proteins) for 30 min. Thereafter, bacteria were washed twice with 1× PBS and resuspended in 1× PBS. Binding of fusion proteins to bacterial surface was measured using CyFlow space flow cytometer (Partec), and geometric mean fluorescence intensity was calculated with FlowJo software (Tree Star). To assess serum FH binding, bacteria were similarly treated, but incubated in 10% NHS and stained with MRC OX-24 anti-human FH conjugated to DyLight 650. To identify bacteria, CellTrace Violet–positive events were gated and further analyzed.

FH6-7/hFc was added to lepirudin (Refludan; Celgene)-treated human blood in different concentrations before infection with indicated amounts of different S. pyogenes strains. Complement activation was blocked at the level of C3 with 20 μM compstatin CP40. Fcγ receptors were blocked with a combination of 10 μg/ml each of anti-human CD16, CD32, and CD64 that were added to whole blood and incubated for 20 min at 37°C before adding bacteria. Following the addition of bacteria, the blood was incubated on an end-over-end shaker at 37°C, 5% CO₂. At indicated time points, 50 μl of the mixture was diluted serially in TBS and plated on blood agar plates. Blood agar plates were incubated overnight at 37°C and 5% CO₂ to enumerate surviving S. pyogenes.

Harvested bacteria were stained with CFSE and subsequently subcultured for 2 h to reduce the CFSE intensity per bacteria. Human blood (200 μl) was infected with 1 × 10⁷ bacteria and incubated for 30 min at 37°C and 5% CO₂. Next, blood samples were treated with human TruStain FcX to block Fc receptors for 10 min on ice prior to addition of Abs. After 30 min incubation, RBCs were lysed using RBC lysis/fixation solution (BioLegend) with subsequent incubation for 15 min at room temperature. After washing with PBS, cells were analyzed using Amnis FlowSight.

The affinity between protein H and FH6-7/hFc was analyzed by surface plasmon resonance using a Biacore 2000 (GE Healthcare). FH-IgG was immobilized on a CM5 sensor chip (GE Healthcare) using standard amino coupling reaction to reach 2500 resonance units. Protein H diluted to 0.07–25 nM concentrations in running buffer (50 mM HEPES [pH 7.8] containing 100 mM NaCl and 0.005% Tween 20) was injected at flow of 30 μl/s for 100 s and the dissociation was then followed for 300 s. The signal from the control surface was subtracted. In all experiments, two consecutive injections of 2 M NaCl, 100 mM HCl were used to remove bound ligands during a regeneration step. The BiaEvaluation 3.0 software (Biacore) was used to determine affinity constants using a Langmuir 1:1 interaction model with drifting baseline. The experiment was performed two independent times.

Platelet aggregation was analyzed from three independent donors, as recommended by the International Society on Thrombosis and Haemostasis (43). Briefly, plasma was purified from healthy volunteers and platelet-rich plasma and platelet-poor plasma were prepared. After mixing platelet-poor plasma and platelet-rich plasma together with 2 μM ADP (positive control), FH6-7/FH, or PBS (negative control), samples were analyzed in a light transmission aggregometer. Additionally, one sample was left untreated before analysis (spontaneous aggregation, additional negative control). OD was measured and plotted as percentage. Area under the curve analysis revealed the total amount of platelet aggregation.

Lepirudin-treated plasma was incubated either with 2 mg/ml zymosan particles, PBS, or 50 μg/ml FH6-7/hFc for 1 h at 37°C with shaking. Particles were removed and supernatants were applied on a 15% SDS-polyacrylamide gel. Proteins were separated by SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and subsequently incubated with mouse anti-human C3a followed by an incubation with goat anti-mouse Ig-HRP. The C3a signal was visualized using ECL (Millipore) and detection with a CCD camera (Chemi-Doc; Bio-Rad). Owing to a very strong signal in the positive control in the initial experiment, the zymosan-treated sample was diluted 10-fold prior to electrophoresis in the experiment shown.

All animals were housed and bred under specific pathogen-free conditions in the animal facility at the University of Massachusetts Medical School (Worcester, MA).

Two hours prior to infection, female and male animals were treated either with 50 μg of FH6-7/hFc or 50 μg of goat IgG i.p. Subsequently, via lateral tail vein injection, animals were infected i.v. with 100 μl of bacterial suspensions in PBS containing 1.4 × 10⁸/ml S. pyogenes AP1. Injections were repeated on days 2, 4, and 6 (50 μg of protein per animal). All animals were closely monitored for signs of disease for up to 8 d; gravely moribund mice were euthanized.

Use of animals in this study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and the Swedish Animal Welfare Act SFS1988:534. Experiments were approved by the Institutional Animal Care and Use Committee at the University of Massachusetts Medical School.

Statistical analysis was performed using GraphPad Prism 6.0 and 7.0b software. To test for significance, we used either a one-way or two-way ANOVA analysis with a Bonferroni, Dunnett, or Tukey posttest or a Mantel–Cox test as indicated. A p value < 0.05 was considered to be significant.

We analyzed binding of FH domains 6 and 7 coupled to human IgG1 Fc (FH6-7/hFc; Fig. 1A) to S. pyogenes AP1 and its isogenic mutants MC25 (lacks the M protein), BM27.6 (lacks protein H), and BMJ71 (lacks both M protein and protein H). FH6-7/hFc bound similarly to the parental wild-type strain AP1 and MC25 (Fig. 1B). Binding to the mutant BM27.6, as measured by fluorescence intensity, was diminished almost 2-fold compared with the wild-type strain. Bacteria lacking both surface virulence factors, M/H proteins, BMJ71 showed binding barely above background levels. These results indicate that FH6-7/hFc binds to both M protein and protein H on S. pyogenes. Prior work has suggested S. pyogenes also binds FH through domains18–20 (44, 45). We observed binding of FH18-20/hFc to the wild-type AP1 and, to a lesser extent, to MC25 (Fig. 1C). Both BM27.6 and BMJ71 did not bind FH18-20/hFc, suggesting that M protein and protein H may both contribute to binding of the C-terminal domains of FH.

Because S. pyogenes is known to bind human, but not mouse, IgG-Fc (39), we employed FH6-7 and FH18-20 fused to mouse IgG2a to determine whether binding of FH6-7/hFc to AP1 is mediated mainly through FH or through human IgG-Fc domains of the fusion proteins tested. As seen with FH6-7/hFc, AP1 and MC25 bound FH6-7/mFc well, whereas BM27.6 and BMJ71 bound reduced amounts or no FH6-7/mFc, respectively, indicating that binding of FH6-7/hFc occurred through the FH domains (Fig. 1D). However, we barely detected any binding of FH18-20/mFc to the bacteria (Fig. 1E). These results indicate that FH18-20/hFc bound almost exclusively through the human IgG-Fc part of the protein, whereas FH6-7 presumably binds via both domains, that is, FH6-7 as well as hIgG1-Fc. Using surface plasmon resonance analysis, we estimated the binding affinity of protein H to FH6-7/hFc to be K_D of 5.2 × 10⁻⁹ M (Fig. 1F) and FH6-7/mFc to be K_D of 3.7 × 10⁻⁹ M (Fig.1G), indicating a strong interaction for both chimeric proteins.

Next, we tested whether FH fusion proteins were able to compete out serum FH on the surface of S. pyogenes. Therefore, we incubated the bacteria in 5% NHS that contained increasing concentrations of FH6-7/hFc (Fig. 2A). Even 2.5 μg/ml FH6-7/hFc was sufficient to completely block FH binding to the bacterial surface of AP1 and MC25. BM27.6 and BMJ71 did not bind any FH from serum. In contrast, FH18-20/hFc did not displace any serum FH from AP1 or MC25 (Fig. 2B). Even high concentrations (150 μg/ml) of FH18-20/hFc had no effect on serum FH binding (data not shown). These data provide further evidence for binding of FH6-7/hFc via the FH domains, in contrast to binding of FH18-20/hFc via the Fc domain.

We hypothesized that FH-IgG binding to S. pyogenes would increase complement attack through enhanced classical pathway activation and opsonization by the Fc part of the chimeric protein, and also by preventing FH-mediated inhibition of complement activation. To confirm our hypothesis, we analyzed C9 deposition on S. pyogenes AP1 from 10% NHS in the presence of increasing amounts of the chimeric proteins; multiple C9 molecules participate in the formation of the membrane attack complex (MAC), which is the final stage in complement activation. We found that FH6-7 significantly increased C9 deposition on AP1 compared with NHS alone (Fig. 2C). In contrast, FH18-20/hFc did not increase C9 deposition on the bacterial surface. As expected, neither fusion protein had any impact on BM27.6 (Fig. 2D), because this mutant does not bind any FH domains. Consistent with uninhibited complement activation on its surface, high MAC formation occurred on the surface of BM27.6 and was not altered by addition of a fusion protein. To determine whether the Fc portion of FH6-7/hFc could activate complement, we coated microtiter plates with the fusion protein, aggregated IgG as a positive control, and α1AT as a negative protein control. Indeed, FH6-7/hFc increased both C3b (Fig. 2E) as well as C4b deposition (Fig. 2F) on the plate similar to aggregated IgG, indicating that FH6-7/hFc activates the alternative and classical pathways. To address the possibility that complement activation in solution may have contributed to complement component deposition on “bystander” bacteria, we incubated plasma for 60 min with either FH6-7/hFc, PBS, or zymosan particles and measured C3a generation as a measure of complement activation. Plasma samples were subjected to Western blot analysis for C3a (Fig. 2G). Zymosan strongly activated complement, resulting in high amounts of C3a production, whereas PBS and FH6-7/FH showed similar, but very weak signals. Taken together, FH6-7/hFc activates complement only when bound to the bacterial surface, but not when in solution.

Initial binding experiments indicated that protein H and M protein were both ligands for FH6-7 (Fig. 1B). To exclude the influence of human IgG-Fc, we used ¹²⁵I-labeled FH6-7/mFc to test binding of the FH domains to protein H, M1, and α1AT (Fig. 3A). FH6-7/mFc bound in a dose dependent-manner to protein H, slightly less to M1, but not to the negative control α1AT.

Because we identified protein H as one of the major ligands for FH6-7/hFc, we sought to identify the binding site on protein H for FH domains 6 and 7 (Fig. 3B). We found that full-length protein H bound most FH6-7/mFc, whereas protein H fragments ∆267–376 and ∆246–376 (see Fig. 3B, insert) still bound significantly more FH6-7/mFc than did the irrelevant protein control α1AT, yet markedly decreased compared with full-length protein H (Fig. 3B). Thus, we concluded that the binding site for FH6-7/mFc is located in domains C (mainly C3) and D of protein H.

Similarly, we used ¹²⁵I-radiolabeled FH to compare binding to M1 and protein H. Similar to FH6-7/mFc, we found that FH binds both protein H as well as M protein (Fig. 3C). Using the protein H fragments, we detected significant binding of ¹²⁵I-FH to full-length protein H and fragments ∆267–376 and ∆246–376 as well as ∆145–376. This binding pattern suggests that full-length FH binds to domains C and D, as well as domain B, of protein H (Fig. 3D).

Unwanted displacement of FH from tissues and cells could be detrimental for the body. Therefore, we asked whether FH6-7/hFc affected hemolysis of erythrocytes. To determine whether FH6-7/hFc augmented classical pathway activation, we sensitized sheep erythrocytes with Abs and subsequently incubated these with FH6-7/hFc. Before addition of serum, cells were washed to assess the function of only erythrocyte-bound fusion protein. We did not detect any difference in hemolytic activity at any tested FH6-7/hFc concentration. Similarly, we used nonsensitized sheep erythrocytes to identify whether FH6-7/hFc could directly sensitize the erythrocytes. Again, there was no increase in hemolysis compared with background hemolysis (Fig. 3E, left side). These data show that FH6-7/hFc had no effect on classical pathway–mediated hemolysis. To analyze the effect of the fusion protein on the alternative pathway, we used rabbit erythrocytes, which were incubated with increasing amounts of FH6-7/hFc. Erythrocytes were washed to remove unbound FH6-7/hFc. As expected, the fusion protein had no influence on the hemolysis of rabbit erythrocytes, suggesting that it does not compete with FH in this experimental system (Fig. 3E, right side).

FH is the major protective molecule on human erythrocytes where CD59 function is blocked, for example with anti-CD59 Abs (46). FH6-7/hFc did not induce hemolysis of anti-CD59–treated human RBCs (Fig. 3F) at any of the tested concentrations. In contrast, FH18-20/Fc, which is known to induce hemolysis, served as a positive control (35). Taken together, our experiments show that FH6-7/hFc does not cause complement-dependent lysis of erythrocytes.

To verify that FH6-7/hFc does not affect coagulation, we performed a platelet aggregation assay (Fig. 3G). Serum from three healthy individuals was analyzed by either adding FH6-7/hFc, PBS as a negative control, or ADP as a positive control. A fourth untreated sample was used to measure spontaneous activation. A comparison of the area under the curves of the different reaction mixtures revealed that FH6-7/hFc did not activate platelets beyond levels seen with PBS and “spontaneous activation” tubes.

We infected human blood from five healthy individuals ex vivo with S. pyogenes AP1 and subsequently stained for professional phagocytes to measure uptake of bacteria. Using Amnis FlowSight technology, we discriminated between bacterial uptake and adhesion (Fig. 4A). We excluded all cells with extracellular bacteria (Fig. 4A, “excluded cells” no. 78784 and no. 81540) and analyzed the CFSE signal in the PMN population. We observed PMNS without phagocytosed bacteria (Fig. 4A, no. 8074 and no. 33332) as well as PMNS that phagocytosed one (no. 16415), two (no. 40855), three (no. 83010), or more than three (no. 84420) bacteria, judged by the CFSE signal originating from the bacteria.

In total, 40.4% of all neutrophils in blood ingested S. pyogenes in the absence of FH6-7/hFc (Fig. 4B, individual donors color matched). Upon addition of the fusion protein, the phagocytosis rate increased in the individual samples between 6.4 and 16.4–52% on average. Addition of cytochalasin D completely blocked phagocytosis and confirmed that we had measured bacterial uptake and not merely adhesion to neutrophils.

To assess the effect of FH6-7/hFc on bacteria during sepsis, we infected human blood with S. pyogenes AP1 in the presence or absence of FH6-7/hFc. Bacterial growth was analyzed during a period of 3 h (Fig. 5A). In the presence of 50 μg/ml FH6-7/hFc (red bars), bacterial burden was reduced by >60% within the first 60 min. During the next 2 h the bacterial numbers remained stable (static) whereas untreated bacteria replicated impressively; after 3 h, the difference between the untreated sample and the samples treated with 50 μg/ml FH6-7/hFc was >1.5 log₁₀. Treatment with 10 μg/ml FH6-7/hFc (gray bars) showed significantly reduced bacterial burden compared with untreated controls, but the decrease was not as pronounced as in samples with 50 μg/ml (red bars). In fact, almost 1 log₁₀ bacterial growth occurred during 3 h in the presence of the low dose. Similarly, blood infected with S. pyogenes AP18 and treated with FH6-7/hFc showed similar reduction in bacterial burden over time. The effect was even more pronounced using a low level (10 μg/ml) of FH6-7/hFc. The difference between treated and untreated blood was ∼2 log₁₀ bacteria/ml at 2 or 3 h after addition of bacteria (Fig. 5B). As expected, bacteria unable to bind FH6-7/hFc were not affected and CFU counts were similar in treated and untreated samples at all time points (Fig. 5C, 5D). This shows that FH6-7/hFc is effective only against bacteria that bind FH via domains 6–7.

We sought to delineate the mechanism whereby FH6-7/hFc increases uptake and killing of group A streptococci. Because phagocytosis is mainly mediated either via Fcγ receptors or complement opsonization, we performed a blood killing assay in the presence of FH6-7/hFc where Fcγ receptor, complement C3 activation, or both were blocked (Fig. 5E). Blocking opsonization with C3 fragments using compstatin led to a significant increase of bacterial survival (∼1 log₁₀ decrease in CFU at each time point tested). Surprisingly, blocking Fcγ receptors alone did not change bacterial blood burden, suggesting that complement activation was essential for FH6-7/hFc-mediated opsonophagocytic killing in whole human blood.

To show in vivo efficiency of the fusion protein, we infected human FH transgenic animals with 1.4 × 10⁷ S. pyogenes AP1. S. pyogenes binds only human, but not mouse, FH. Therefore, use of human FH transgenic mice provided a complement-inhibiting barrier that FH6-7/hFc may encounter in humans. Animals treated with 50 μg of FH6-7/hFc every other day showed significantly reduced mortality compared with the control group receiving control goat IgG (Fig. 6). Four out of ten animals treated with the fusion protein survived the infection, whereas all animals receiving control goat IgG died within 5 d. These data provide evidence that FH6-7/hFc reduces bacterial burden in blood and increases survival of S. pyogenes–infected animals.

Taken together, our data suggest that FH6-7/hFc blocks FH binding to bacteria, leading to increased complement deposition, enhanced phagocytosis, and impaired bacterial survival (Fig. 7).

In this study, we demonstrate efficacy of a FH-IgG fusion protein against S. pyogenes infections, thus expanding their utility against Gram-positive bacteria that bind FH. FH-Fc chimeric proteins could be further developed as a new class of antibacterial proteins, which activate complement to opsonize their targets for elimination by phagocytes or, in case of Gram-negative bacteria, directly lyse them by MAC.

FH6-7/hFc has already been proven efficacious against N. meningitidis (34) and H. influenzae (33), whereas a derivative of FH18-20/Fc that contains a D to G mutation at position 1119 in domain 19 was effective against N. gonorrhoeae (35). Of note, all three bacterial species belong to the group of Gram-negative bacteria, which are generally susceptible to complement-mediated lysis (17, 47). To achieve MAC-mediated lysis of Gram-negative bacteria, the chimeric FH-Fc needs to bind to bacteria to displace FH from the bacterial surface and or activate the classical pathway through Fc. However, complement alone cannot directly lyse Gram-positive bacteria due to their thick cell wall (48). Instead, complement-mediated opsonization of the bacteria with iC3b is crucial for efficient phagocytosis. We showed that the chimeric protein bound to the bacterial surface and competed out FH, increased complement activation, led to a significant increase of phagocytosis by PMNs, reduced bacterial burden in human blood ex vivo, and significantly improved survival of mice in a S. pyogenes sepsis model.

Many pathogens have developed the ability to recruit complement inhibitors such as FH and C4BP to their surface to evade complement (reviewed in Refs. 20, 32). FH and C4BP tightly regulate complement activation by accelerating the decay of the C3 convertase and serving as cofactors for factor I–mediated inactivation of C3b (49–52). Complement activation is dampened on pathogens “coated” with these inhibitors and thus are protected from opsonophagocytosis (44). Blocking the binding of complement inhibitors renders the bacteria again susceptible to complement and would allow the immune system to eliminate the invaders. FH-IgG chimeric protein acts by competing out serum FH bound to those bacteria and also activates the classical pathway through its Fc region (33–35).

Similar to at least 10 other bacterial species (26), most strains of S. pyogenes recruit serum FH through domains 6 and 7 to their surface to protect themselves from complement attack (53, 54). The regions in FH that interact with microbes (domains 5–7 and 19–20) are distinct from the domains responsible for complement inhibition (domains 1–4). Thus, FH6-7/hFc lacks any complement inhibiting activity, prevents binding of functional host FH to the pathogen, and also activates complement through Fc. Given that several pathogens bind FH through a similar region in FH, it raises the possibility that FH6-7/hFc may possess activity against several other microbes (32).

To our surprise, we found that in addition to protein H, the M1 protein from S. pyogenes AP1 also bound FH6-7-IgG. Protein H, but not the M1 protein, has been described to bind FH (13, 55, 56). Interestingly, FH seems to have more than one binding site: domains C and D, similar to FH6-7/Fc and presumably the N-terminal part of domain B. Strain AP1 used in this study expresses both protein H and M1, whereas the previous study employed a different M1-expressing strain, S. pyogenes SF370. It is possible that differences in the M proteins between the strains may account for differences in FH binding to these S. pyogenes strains.

S. pyogenes is known to bind human, but not mouse, IgG (39, 57). Thus, we used FH/Fc constructs that contained mouse IgG-Fc in this study to show that binding of FH6-7/hFc occurred specifically via the FH domains. In contrast, FH18-20/hFc bound S. pyogenes exclusively through human IgG domains, thus explaining why FH18-20 did not compete out serum FH and did not enhance complement activation on the bacteria. Although we cannot exclude the possibility that a proportion of FH6-7/hFc may also bind to the bacteria via the human IgG part, we have shown that even a low amount (∼50 nM) of FH6-7/hFc is able to effectively compete out FH completely.

Another advantage of FH6-7/hFc is that drug resistance, if it were to occur, would lead to a loss of binding of endogenous FH and FH-like protein 1 that both possess complement inhibiting function and thereby protect the bacteria. Loss of FH/FH-like protein 1 binding would lead to a considerable decrease of bacterial fitness in vivo because the resulting increase in C3 fragment deposition would lead to bacterial clearance by phagocytes (13, 53).

In comparison with Gram-negative organisms, Gram-positive bacteria cannot be lysed through complement-mediated pore formation, although MAC can still be detected on their surface (58). Although the role of MAC associated with Gram-positive bacteria is unclear, a recent study showed that MAC can be transferred from the surface of an opsonized particle to the macrophage plasma membrane and initiate the NLRP3 inflammasome, resulting in caspase-1 activation and IL-1β and IL-18 secretion (59).

Therefore, the mode of action of FH6-7/hFc on S. pyogenes must differ from the previously described mechanisms for Neisseria and Haemophilus. If bacteria cannot be directly lysed by complement, professional phagocytes must clear the microbes employing either extracellular killing mechanisms or intracellular degradation after phagocytosis. We showed that FH/Fc increased opsonization and following opsonophagocytosis. This was presumably caused on the one hand by displacement of FH and concomitant loss of protection from complement deposition, and on the other hand the additional activation of the classical complement pathway through the IgG1-Fc domains of our chimeric protein (60). In addition to C3b and iC3b, the Fc domain of the IgG1 backbone engages the Fcγ receptor on professional phagocytes (61). Fcγ receptor engagement alone does not mediate killing of S. pyogenes, whereas complement-driven opsonization is the most important pathway to clear the bacteria (62). Using compstatin CP40 we confirmed this result in untreated human full blood as well as in the presence of FH6-7/Fc. Despite FH6-7/Fc binding to bacteria via FH domains and presenting the hIgG1-Fc, blocking Fcγ receptors (CD16, CD32, and CD64) had no influence on S. pyogenes killing. Compstatin in turn promoted bacterial survival by blocking complement activation at the level of C3. Of note, although compstatin prevents C3b deposition on the bacteria, at the same time it allows for increased C4b deposition (63), which may serve as an opsonin by binding CR1 (64). The potential relevance of this enhanced C4b deposition is still not clear. Noncomplement bactericidal effects of serum, such as transferrin, antimicrobial peptides, and enzymes, also contribute to bacterial killing. Further work is needed to delineate the exact mechanism of action of FH6-7/Fc on Gram-positive bacteria. Taken together, the FH6-7/hFc mediates opsonophagocytosis through two mechanisms: 1) activating complement via C1q–IgG1 interaction, and 2) blocking FH-dependent complement inhibition (see figure 2 in Ref. 26).

It is urgent to develop novel approaches to cope with antibiotic-resistant bacterial infections, which pose a major threat to human health globally. The Centers for Disease Control as well as the World Health Organization have listed 15 pathogens as concerning, serious, or urgent threats that need close monitoring, preventive activities, or even urgent and aggressive actions to counteract their antibiotic resistance threat (29, 65). Among those 15 different bacterial species, 9 are known to bind FH, of which at least 7 employ domains 5–7 for binding (26, 32). Because FH6-7/hFc especially targets those binding sites, we think that the chimeric protein could be a valuable adjunctive immunotherapeutic against antibiotic-resistant bacterial pathogens. Although FH/Fc fusion proteins have shown promise in preclinical models, we acknowledge that further work needs to be undertaken to establish their safety and stability for use in humans.

We thank Nancy Nowak and Bo Zhang for expert help with breeding and caring for mice and the Center for Thrombosis and Haemostasis, Skåne University Hospital, Malmö, Sweden for help with the coagulation assays.

S.R. and J.S. are inventors on a patent application in the area of FH/Fc therapeutics. J.D.L. is the inventor of patents and/or patent applications that describe the use of complement inhibitors for therapeutic purposes; the founder of Amyndas Pharmaceuticals, which is developing complement inhibitors (i.e., next generation compstatins) for clinical applications; and the inventor of the compstatin technology licensed to Apellis Pharmaceuticals (i.e., 4(1MeW)7W/POT-4/APL-1 and its PEGylated derivatives). The other authors have no financial conflicts of interest.